From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.linuxfoundation.org ([140.211.169.12]:55646 "EHLO mail.linuxfoundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751967AbdEAV2B (ORCPT ); Mon, 1 May 2017 17:28:01 -0400 From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman , stable@vger.kernel.org, Jamal Hadi Salim , Cong Wang , "David S. Miller" , Julia Lawall Subject: [PATCH 4.4 11/43] net_sched: close another race condition in tcf_mirred_release() Date: Mon, 1 May 2017 14:27:11 -0700 Message-Id: <20170501212559.988601385@linuxfoundation.org> In-Reply-To: <20170501212559.546911128@linuxfoundation.org> References: <20170501212559.546911128@linuxfoundation.org> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Sender: stable-owner@vger.kernel.org List-ID: 4.4-stable review patch. If anyone has any objections, please let me know. ------------------ From: WANG Cong commit dc327f8931cb9d66191f489eb9a852fc04530546 upstream. We saw the following extra refcount release on veth device: kernel: [7957821.463992] unregister_netdevice: waiting for mesos50284 to become free. Usage count = -1 Since we heavily use mirred action to redirect packets to veth, I think this is caused by the following race condition: CPU0: tcf_mirred_release(): (in RCU callback) struct net_device *dev = rcu_dereference_protected(m->tcfm_dev, 1); CPU1: mirred_device_event(): spin_lock_bh(&mirred_list_lock); list_for_each_entry(m, &mirred_list, tcfm_list) { if (rcu_access_pointer(m->tcfm_dev) == dev) { dev_put(dev); /* Note : no rcu grace period necessary, as * net_device are already rcu protected. */ RCU_INIT_POINTER(m->tcfm_dev, NULL); } } spin_unlock_bh(&mirred_list_lock); CPU0: tcf_mirred_release(): spin_lock_bh(&mirred_list_lock); list_del(&m->tcfm_list); spin_unlock_bh(&mirred_list_lock); if (dev) // <======== Stil refers to the old m->tcfm_dev dev_put(dev); // <======== dev_put() is called on it again The action init code path is good because it is impossible to modify an action that is being removed. So, fix this by moving everything under the spinlock. Fixes: 2ee22a90c7af ("net_sched: act_mirred: remove spinlock in fast path") Fixes: 6bd00b850635 ("act_mirred: fix a race condition on mirred_list") Cc: Jamal Hadi Salim Signed-off-by: Cong Wang Acked-by: Jamal Hadi Salim Signed-off-by: David S. Miller Cc: Julia Lawall Signed-off-by: Greg Kroah-Hartman --- net/sched/act_mirred.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) --- a/net/sched/act_mirred.c +++ b/net/sched/act_mirred.c @@ -36,14 +36,15 @@ static DEFINE_SPINLOCK(mirred_list_lock) static void tcf_mirred_release(struct tc_action *a, int bind) { struct tcf_mirred *m = to_mirred(a); - struct net_device *dev = rcu_dereference_protected(m->tcfm_dev, 1); + struct net_device *dev; /* We could be called either in a RCU callback or with RTNL lock held. */ spin_lock_bh(&mirred_list_lock); list_del(&m->tcfm_list); - spin_unlock_bh(&mirred_list_lock); + dev = rcu_dereference_protected(m->tcfm_dev, 1); if (dev) dev_put(dev); + spin_unlock_bh(&mirred_list_lock); } static const struct nla_policy mirred_policy[TCA_MIRRED_MAX + 1] = {